W6 L2 Data interpretation: Absolute vs Relative Risk and Misleading Axes

Absolute vs Relative Risk – Core Notes

Key Definitions

• Absolute Risk (AR): actual probability of an event in a group.

• Relative Risk (RR): compares risk between groups.

• Absolute Risk Difference (ARD): difference in absolute risk between groups.

• Relative Risk Increase (RRI): proportional increase above control.

Interpretation:

• RR > 1 → higher risk in exposed.

• RR = 1 → no difference.

• RR < 1 → protective effect.

Worked Examples

1. Breakfast Skipping & Mortality

• AR (no breakfast) = 0.64%

• AR (skip breakfast) = 0.73%

• RR = 1.14 → 14% higher risk

• ARD = 0.09% (tiny absolute change)
  ➡ RR sounds dramatic, AR shows it’s very small in real terms.

2. Post-Vaccination Myocarditis

• AR = 380 per 1,000,000 (0.038%)

• RR ≈ 2800 compared to baseline
  ➡ Huge relative increase, but <0.1% absolute risk → very rare.

3. Sleep & Dementia

• ARD = 0.1%

• RR = 1.3 (30% higher risk)
  ➡ Relative effect looks big, absolute effect is small.

Data Visualization Pitfalls

• Truncated y-axes: exaggerate small differences.

• Truncated x-axes / cherry-picking timeframe: distort trends.

• Selective reporting: omits baseline risks, misleads.

Best Practice in Reporting

• Always give both AR and RR (plus ARD).

• State baseline risks.

• Show complete axes and full time periods.

• Contextualize: small AR but high RR ≠ large population risk.

• Be transparent about definitions, sources, and limitations.

Big Takeaways

✅ AR = actual probability.
✅ RR = comparative likelihood.
✅ ARD = real-world difference (percent points).
✅ RRI = proportional increase.
✅ Both AR & RR needed for balanced communication.
✅ Visualization choices (axes, timeframes) shape perception → avoid misleading displays.